Texture Evaluation of Carpets Using Image Analysis

Abstract

This paper explores in detail new image enhancement techniques for assessing tuft geometry and concomitant appearance changes in new and worn carpets. Using im proved algorithms and more advanced tuft geometry, saxony tufted carpets of nylon and polyester BCF yams are capable of being extensively and reliably characterized as to texture and changes in appearance with laboratory or floor wear. While still using a relatively low resolution grey level imaging system, the newly developed al gorithms demonstrate how tuft size distribution, number of tufts, evenness of tuft spatial distribution, and aspect ratio (shape factor) of tufts can be objectively measured. Statistical analysis confirms that customary wear levels, and their visual appearance change in carpet, can be separated and correctly grouped by these new techniques. Some limitations of these and other texture-sensitive image analysis techniques are also discussed.